The closest I came was when I had some papers on my desk that were given out for free somewhere and somebody informed me that that was used mainly for smoking pot. I had no idea.
It's almost an opposite of Hanlon's razor: "Attribute to drugs what you could attribute to lack of sleep (or similar)". That plus physical tells are the biggest indicators. In the end I wouldn't say that could give a definite evaluation if someone is using drugs or not in most cases.